Professional Documents
Culture Documents
Social Processes in Children's Learning: Paul Light and Karen Littleton
Social Processes in Children's Learning: Paul Light and Karen Littleton
Social Processes in Children's Learning: Paul Light and Karen Littleton
Children’s Learning
This is the Wrst book to bring together the extensive modern evidence for
innate imitation in babies. Modern research has shown imitation to be a
natural mechanism of learning and communication which deserves to be
at centre stage in developmental psychology. Yet the very possibility of
imitation in newborn humans has had a controversial history. DeWning
imitation has proved to be far from straightforward and scientiWc evi-
dence for its existence in neonates is only now becoming accepted,
despite more than a century of enquiry. In this book, some of the world’s
foremost researchers on imitation and intellectual development review
evidence for imitation in newborn babies. They discuss the development
of imitation in infancy, in both normal and atypical populations and in
comparison with other primate species, stressing the fundamental im-
portance of imitation in human development, as a foundation of com-
munication and a precursor to symbolic processes.
Learning to Read and Write: A Cross-Linguistic
Perspective
Margaret Harris and Giyoo Hatano
For many years, the development of theories about the way children
learn to read and write was dominated by studies of English-speaking
populations. As we have learned more about the way that children learn
to read and write other scripts – whether they have less regularity in their
grapheme–phoneme correspondences or do not make use of alphabetic
symbols in all – it has become clear that many of the diYculties that
confront children learning to read and write English speciWcally are less
evident, or even non-existent, in other populations. At the same time,
some aspects of learning to read and write are very similar across scripts.
The unique cross-linguistic perspective oVered in this book, including
chapters on Japanese, Greek and the Scandinavian languages as well as
English, shows how the processes of learning to read and spell are
aVected by the characteristics of the writing system that children are
learning to master.
Children’s Understanding of Biology and Health
Michael Siegal and Candida Peterson
This book uses new research and theory to present the Wrst state-of-the-
art account of children’s understanding of biology and health. The
international team of distinguished contributors views children’s under-
standing in these areas to be to some extent adaptive to their well-being
and survival and uses evidence collected through a variety of diVerent
techniques to consider whether young children are capable of basic
theorising and understanding of health and illness. Topics ranging from
babies to the elderly including birth, death, contamination and con-
tagion, food and pain are examined and close links between research and
practice are made with obvious attendant beneWts in terms of education
and communication.
XXXXXX
Series Editors
George Butterworth (General Editor), University of Sussex, UK
Giyoo Hatano, Keio University, Tokyo, Japan
Kurt W. Fischer, Harvard University, USA
Advisory Board
Patricia M. GreenWeld, University of California, Los Angeles, USA
Paul Harris, University of Oxford, UK
Daniel Stern, University of Geneva, Switzerland
Esther Thelen, Indiana University, USA
The aim of this series is to provide a scholarly forum for current theoretical and
empirical issues in cognitive and perceptual development. As the twentieth cen-
tury draws to a close, the Weld is no longer dominated by monolithic theories.
Contemporary explanations build on the combined inXuences of biological,
cultural, contextual and ecological factors in well-deWned research domains. In
the Weld of cognitive devlopment, cultural and situational factors are widely
recognised as inXuencing the emergence and forms of reasoning in children. In
perceptual development, the Weld has moved beyond the oposition of ‘innate’ and
‘acquired’ to suggest a continuous role for perception in the acquistion of knowl-
edge. These approaches and issues will all be reXected in the series which will also
address such important research themes as the indissociable link betweeen per-
ception and action in the developing motor system, the relationship between
preceptual and cognitive development to modern ideas on the development of the
brain, the signiWcance of developmental processes themselves, dynamic systems
theory and contemporary work in the psychodynamic tradition, especially as it
relates to the foundations of self-knowledge.
1. Imitation in Infancy
Jacqueline Nadel and George Butterworth
Forthcoming titles
Nobuo Masataka
The Onset of Language
XXXXXX
http://www.cambridge.org
References 101
Index 116
ix
Figures
6.3 The mean recognition score for the girls and boys for
the two modes of task presentation 83
6.4 The perceptual-motor skills task 85
6.5 The mean ‘hits’ for the girls and boys in the test and
game conditions 86
XXXXXX
The writing of this volume was made possible by Economic and Social
Research Council Fellowship awards to both authors (Award numbers:
H52427501595 and H52427000994). Carole Kershaw and Lynda Pres-
ton gave invaluable help in preparing the manuscript for publication. Ian
Wallhead and Simeon Yates kindly reproduced the Wgures. The empirical
research on which the volume is based was supported by project grants
from the ESRC, the Leverhulme Trust and the NuYeld Foundation.
Other members of the research teams included Martin Glachan, Chris-
topher Colbourn, Richard Joiner, Peter Barnes, David Messer, Agnes
Blaye, Maria Silvia Barbieri and Annerieke Oosterwegel. The research
also involved many schools and many hundreds of children, without
whose willing and unrewarded cooperation none of the research could
have been conducted. Our sincere thanks to them all.
Figures 4.1 and 4.2 are reprinted with permission from Springer-Verlag.
These Wgures originally appeared in Computer Supported Collaborative
Learning (1995), edited by Claire O’Malley.
xviii
XXXXXX
Introduction
This chapter will set the scene for those that follow by going back to the
1970s to examine some of the strands of theory and empirical research
which converged around the question of when and how peer interaction
can facilitate children’s understanding and learning. Starting with social
learning theory, we shall develop a focus on the concept of socio-cognitive
conXict as an engine of mental development, a concept that owes its
origins to Piaget, via Doise and colleagues. The Piagetian origins of this
idea are reXected in the selection of tasks in these early studies, and this in
turn gives rise to certain problems in terms of the interpretation of some
of the experimental Wndings. This consideration will take us on a slight
detour in the course of this chapter, raising issues to be returned to later.
The latter part of the chapter will be given over to an account of the series
of experimental studies of peer facilitation of children’s problem solving
which marked our own initial engagement with this Weld of research.
who had worked in pairs in the second session than for those who had
worked alone.
Large diVerences between the children in terms of their pre-test levels
were associated with less progress than small diVerences in initial ability
(Doise and Mugny, 1984). Even children who showed identical levels of
pre-test performance could beneWt from working together if steps were
taken to ensure that they would come up with conXicting responses.
Thus, with the village task, Doise and colleagues arranged for such
children to occupy diVerent spatial positions relative to the array. This
meant that their ‘egocentric’ responses would ensure that they came into
conXict about where to place the buildings, even though they were both
reasoning in the same way. Children paired under these conditions made
more gains than similar children working on the tasks on their own.
Doise and colleagues interpreted their Wndings in terms of socio-
cognitive conXict. The children in the pair or small group conditions
found themselves confronted with solutions which conXicted with their
own. This conXict, and the socially engendered need to resolve it,
prompts the children to re-examine their own initial responses, and may
lead them to recognise a higher order solution to the problem which
resolves the apparent conXict (Mugny, Perret-Clermont and Doise,
1981). For this to occur, it is necessary that the children’s initial solutions
diVer, but it is not necessary for any of them to be more advanced than the
others, or for any of them to be correct. The real ‘ratchet’ driving the
process is that resolution of children’s partial or centred solutions can in
the end only be found by adopting a higher level, more decentred sol-
ution, thus ensuring cognitive progress.
Perret-Clermont (1980) used essentially the same three-stage experi-
mental design in a series of experiments on peer facilitation of conserva-
tion judgements. Pre-tests were individual, and involved a range of stan-
dard Piagetian assessments of children’s understanding of conservation
of quantity. Various arrangements were tried out for the children assigned
to the social interaction condition in the second session. One which
proved eVective was to assign two ‘conservers’ and one ‘non-conserver’ to
work together. The non-conserver was then given a task such as sharing
out between the children by pouring from a jug into three diVerent
shaped glasses. The session went on until all the children agreed that they
each had the same quantity. The non-conserving children exposed to this
kind of interaction went on to show signiWcant pre- to post-test gains on
standard tests of conservation of liquid quantity.
With conservation, as with the village task, progress could result even
from interaction between two non-conservers, provided that they gener-
ated diVering initial judgements. These studies further highlighted the
Peer interaction and learning 5
children, who were tested in pairs. The game was such that it was
important for quantities to be equal, but the chip in the beaker would
make the game dangerous; thus the need to Wnd another container. A
substantial majority of Wve and six-year-olds judged that the resulting
transformation did not aVect the quantities involved, whereas almost all
children of this age fail on the standard version of the same task.
This result has proved replicable (e.g. Miller, 1982), but the interpreta-
tion put on it at the time may well be the wrong one. That interpretation
stressed the fact that the transformation was made to seem incidental to
the proceedings, rather than central to them. Later research using diVer-
ent designs (Roazzi and Bryant, 1992) has failed to Wnd signiWcant eVects
for such ‘incidentality’. On the other hand, our own subsequent studies
have shown that setting the conservation task in the context of a game is
suYcient by itself to improve children’s performance.
For example, Light, Gorsuch and Newman (1987) presented pairs of
Wve and six-year-olds with a heap of dried peas for them to divide into two
equal heaps. These were then put into two rather diVerently shaped
containers, and the children were asked whether the quantities remained
equal. For some of them, all this was done in the context of setting up a
game in which they were going to compete with one another to move their
peas as fast as they could to a target container, using a straw. The children
who encountered the conservation questions in this context were much
more likely to respond correctly than those who encountered the same
transformation and the same questions without the game setting. So it
seems that when children are working in pairs and anticipating a competi-
tive game, they construe the conservation test procedures diVerently. In
this situation they remain resolutely attentive to the quantities involved,
and are not readily distracted by the appearances.
Things are actually rather more complicated. It can be argued (Light,
1986) that in these modiWed versions of the conservation test the children
are really just complying with the apparent wishes and expectations of the
experimenter. Just as the standard versions of these tasks might lead
children toward the wrong response, these versions may in various ways
cue the correct response, even in children with no Wrm grasp of conserva-
tion. Indeed we have been able to show that similar modiWcations can
elicit ‘conserving’ responses even in situations where the quantity in
question is not in fact conserved (Light and Gilmour, 1983).
For present purposes, however, this is not really important. What is
important is that serious question marks were appearing about the valid-
ity of these kinds of Piagetian tests, and the stability of children’s re-
sponses to them. More particularly, there are clear suggestions from this
literature that children in pairs or small groups may well interpret given
Peer interaction and learning 7
tasks and questions diVerently from the way in which the same tasks and
questions would be understood by individuals. Improvements in per-
formance, at the time and subsequently, may reXect this altered under-
standing of the questions at least as much as it reXects any socio-cognitive
conXict arising from diVerent points of view within the group.
One thing that makes this alternative point of view attractive is that
peer facilitation processes sometimes seem to work just too well. Thus, for
example, Perret-Clermont (1980) found that social class diVerences in
children’s pre-test performance on conservation tasks were typically
large, but after a session of interaction of the kind described earlier, the
diVerences according to class disappeared. Similar Wndings have been
reported for urban–rural diVerences, and ethnic diVerences (Light and
Perret-Clermont, 1991).
If the pre-test diVerences were genuinely reXective of diVerences in the
children’s levels of achievement in this crucial area of cognitive develop-
ment, is it really conceivable that such diVerences could be wiped out by a
single session of ten to Wfteen minutes of interaction? It seems much more
plausible that the initial diVerences reXect diVerences in children’s under-
standing of the meaning and reference of the conservation questions, and
indeed there is some independent evidence that this is the case (Grossen,
1988).
It would seem that what disambiguates the questions for the children is
the experience of sharing in the paired or group session, leading Light and
Perret-Clermont to conclude that: ‘the eYcacy of the peer interaction
procedure arises not (or not only) from the socio-cognitive conXict mech-
anism . . . but from the introduction of a norm of equality which serves to
support correct responses, which are then carried over to the individual
post-test’ (1991, p.145).
This issue of how social norms inXuence cognitive functioning surfaced
in a number of areas of research in the 1980s, not least in Doise’s own
work on the inXuence of ‘social marking’. However, we shall leave the
further exploration of this issue for a later chapter. In the remaining
sections of this chapter, we shall turn from Piaget to Vygotsky to Wnd a
rather diVerent set of ‘starting points’ for research on social (and more
particularly peer) processes in learning.
For example, Forman and Cazden (1985) studied joint problem solv-
ing by pairs of nine- to fourteen-year-olds. Their problem-solving tasks
included Piaget’s chemical combinations task (Inhelder and Piaget,
1958) and a task involving predicting the shape of the shadows of vari-
ous objects, when lit from various positions. Their observations of ex-
changes between the children highlighted the importance of well co-
ordinated joint activity in, for example, setting up the apparatus and
planning the experiments. Establishing shared goals and building a
shared understanding of the task are just the kinds of ‘co-constructive’
activities that a Vygotskian analysis might lead one to look for in eVec-
tive peer interaction. On the other hand, in the interpretation of the
results of the experiments, Forman and Cazden saw clear indications
that diVerences of interpretation, defended and argued over, could be
productive in much the way that a Piagetian analysis would lead one
to expect.
Both theoretical approaches may thus have something to oVer, in
drawing attention to diVerent aspects of productive interaction. While a
Piagetian approach highlights what being confronted with a socio-cogni-
tive conXict can do for individual cognitive development, a Vygotskian
approach highlights the way in which a shared understanding can be
arrived at through a process which might be termed ‘mutual construc-
tion’. We shall return to this issue in more detail in subsequent chapters,
but here we want to turn to a rather wider reading of what Vygotsky might
have to oVer to an understanding of social processes in learning.
Introduction
Our own interest in conducting empirical research on peer interaction as
a facilitator of children’s problem solving was raised mainly by exposure
to the neo-Piagetian research of Doise and colleagues, described in the
previous chapter. However, the doubts raised earlier about the validity of
Piagetian procedures for assessing conservation were also being raised by
our own and others’ research in relation to a wide range of other Piagetian
tests. These included the spatial perspective taking tasks which had been
so extensively used by Doise and colleagues in their peer interaction
studies (Donaldson, 1978; Light and Nix, 1983).
Since our interest was the more general question of when, if at all, ‘two
heads are better than one’, it seemed sensible to move beyond the
particular case of children’s mastery of ‘concrete operational reasoning’
to consider a wider range of ages and types of task. Nonetheless, as will be
apparent from the account which follows, the basic three-stage experi-
mental design which had been adopted by Doise and others was carried
over fairly intact into these studies.
tiles are to be moved, as in Figure 2.1, then the player has to move the
smallest to (C), the middle tile to (B), the smallest to (B), the largest to
(C), the smallest to (A), the middle tile to (C) and Wnally the smallest to
(C), a total of seven moves. The solution path thus involves a series of
‘detours’, which form nested subroutines.
We conducted two experiments with this task (Glachan and Light,
1982; Light and Glachan, 1985) using a version very like that shown in
Figure 2.1. In essence, we simply wanted to know whether two
children, performing the task together over a series of trials, would learn
how to do it better than a single child would. As far as the children’s
learning outcomes were concerned, would ‘two heads be better than
one’?
Our Wrst study was conducted with twenty-eight eight-year-olds. All
were given a Wrst session with the task (a ‘pre-test’) on their own, in
which they completed three trials with two-tile towers (which are very
easy) and then three trials with three-tile towers. Their pre-test scores
were based on the three-tile trials, all of which began with the tiles on one
of the outer pegs.
Two weeks later, the children all had a second session, which we can
call the practice session. This time twelve were assigned (at random) to
work alone, while the remaining sixteen children worked in eight pairs,
again with assignment being random except that all were same-sex pairs.
16 Social processes in children’s learning
On this occasion they had eight trials on the Tower of Hanoi, all with
three tiles and all beginning on the centre peg, with the ‘goal’ peg
alternating from trial to trial. They were told to try to solve the problem in
the fewest possible moves, and that the minimum possible number of
moves was seven. The children working in pairs sat opposite one another,
at opposite sides of the puzzle. They were told that tile moving must be
done by both of them together, one holding each handle. A week later, all
children had a Wnal post-test session which was individual and identical to
the pre-test.
Analysis of strategies for solving the task suggested a number of more
or less eYcient strategies, taking seven, nine, eleven and thirteen moves.
Solutions taking eight, ten, twelve and fourteen moves reXected use of
one of these strategies combined with a corrected error at some point.
Solutions taking more than fourteen moves all seemed to contain hap-
hazard sequences of moves, and were treated as ‘non-strategic’.
Children were classiWed by their predominant solution patterns at
pre-test.
The study revealed that there was signiWcantly greater improvement
from pre-test to post-test for those children who had been paired for the
practice session than for those who had worked alone. Closer analysis
indicated that this diVerential beneWt from working in pairs applied only
to those children who showed some sort of strategic approach to the
problem at the pre-test stage. Since children were assigned to pairs at
random, most of the pairs comprised children with diVering pre-test
performances. We looked at those children who had been the ‘better’
member of their pair, in terms of pre-test performance, and matched
them up with children from the individual condition with similar pre-test
performance. Those who had worked in pairs still showed signiWcantly
greater pre- to post-test improvement, indicating that in at least some
circumstances the experience of working in pairs can be advantageous
even if one’s partner is less able than oneself.
This time, pre- and post-test trials included both the ‘centre-peg start’
trials used in the second session and the ‘end-peg start’ trials used in the
previous pre- and post-tests. All types of second session were equally
eVective in terms of producing good post-test performances on the
‘centre-peg start’ trials actually used in the second session. However, the
structured interaction condition was signiWcantly superior to the other
conditions as regards pre- to post-test improvement on the ‘end-peg start’
trials. In fact it seemed that only children in the structured interaction
condition were able to extend what they had learned to related problems.
These results lend themselves quite well to interpretation in terms of
the productive potential of cognitive conXict. The fact that children
lacking any strategic approach to the task were relatively unaVected by the
various manipulations is consistent with the idea that one has to have a
viewpoint of one’s own in order to experience a conXict between this
viewpoint and that of a partner. In order to experience conXict, children
must be pursuing some kind of plan or strategy of their own.
The ineVectiveness of direct coaching (which was done by modelling
solutions rather than by explanations) in producing generalised improve-
ment may be seen in terms of the children’s lack of opportunity to
experiment and resolve conXicts between the modelled solution and their
own previous strategy.
The superiority of the ‘structured’ over the ‘unstructured’ interaction
condition is illuminated by the observation (from videotape analysis) that
while both children appeared to be genuinely engaged in all moves in the
structured condition, almost 90 per cent of moves in the unstructured
condition appeared to be determined and carried out entirely by one
child. There was little discussion and almost no indication of joint deci-
sion making. Moreover, many of these pairs were marked by a high
degree of dominance. In the majority of the unstructured pairs, one of the
children determined at least twice as many moves as the other. In this
situation the dominant individual would have experienced little conXict,
while the submissive individual would have little opportunity to resolve
any conXict experienced.
These Wrst two studies, then, attest to the potential value of peer
interaction in problem solving, using a task remote from the Piagetian
repertoire. They indicated that the productivity of this form of interaction
depends on the parties to the interaction having deWnite ideas of their own
about how to approach the task, and having an opportunity to express
these. They make clear that the presentation of a correct solution to the
problem is neither a necessary nor a suYcient condition to produce
progress. Under some circumstances at least, it seemed, two wrongs
could indeed make a right.
18 Social processes in children’s learning
practice session (session two), the pairs did rather better than the individ-
uals, with nothing to choose between the two types of pairing. However,
at post-test, fully correct (seven move) solutions were signiWcantly and
substantially more frequent in the structured condition than in either of
the others, though only for the trials which reproduced the moves prac-
tised in the earlier sessions.
It seems, then, that what the children are eVectively learning in the
structured condition is not a generalised solution strategy, but rather a
particular sequence of moves. Nonetheless, the fact that only the struc-
tured condition led to any detectable superiority over individual practice
conWrms the indications from the previous study that it is not simply
exposure to solutions that diVer from your own, but actually being
forced to engage with them in some way, which fosters improved per-
formance.
Children working collaboratively on the Tower of Hanoi task, whether
using the ‘physical’ or the computer version, did not typically talk to one
another very much. This is of some interest in that it suggests that talk is
not the ‘be all and end all’ of collaborative learning, an issue to which we
shall return in later chapters. But at the time we saw the dearth of
task-related talk as a drawback, inasmuch as it made it diYcult to exam-
ine relationships between the quality of interaction in a particular pair and
20 Social processes in children’s learning
the progress made by the members of that pair. Talk is relatively easy to
‘collect’, and more amenable to analysis than non-verbal aspects of
interaction. For this reason, amongst others, we extended our research
programme to a rather diVerent type of problem, namely a code-breaking
problem based on the pegboard game ‘Mastermind’.
mean of 8 entries in the case of the older children and 11 entries in the
case of the younger children to solve each problem. Those who had
worked alone throughout required 11 and 15 entries respectively. Thus
the seven year olds who had worked in pairs were as good at the task
at post-test as the twelve- to thirteen-year-olds who had worked on
their own.
There were also promising indications from the transcribed video-
recordings that the nature and extent of the talk that went on between the
children in the paired sessions was correlated with these children’s post-
test performances. However, the lack of any adequate measure of pre-test
performance in this study meant that it was not possible to establish a
causal relationship; it might simply have been a reXection of children with
a better grasp of the task talking more. An adequate pre-test required us
to be able to specify simpler, introductory two-digit ‘code’ sequences, and
then to control the sequences selected to ensure comparability across
participants. We achieved these aims by creating our own microcom-
puter-based version of the task for a second study.
Once again, eighty children of around twelve years of age took part, and
the conditions were just the same as in the previous study. The results
showed even more clearly than in the previous study that neither pairing
nor justiWcation had much impact on their own, but when combined in
the ‘pairs plus justiWcation’ condition the result was signiWcantly better.
Average pre- to post-test improvement in the ‘pairs plus justiWcation’
condition was in fact more than twice as great as for any of the other
conditions (Light and Foot, 1987).
Overview
As will be apparent, the studies described in this chapter involved a fairly
arbitrary choice of tasks. Most were tasks which had been used by other
students of problem solving, and were essentially intellectual rather than
sensorimotor problems. The Tower of Hanoi task demanded forward
planning, while Mastermind required a systematic use of feedback and a
recognition of the signiWcance of negative evidence. The balance beam
comes nearest to the kinds of Piagetian task discussed earlier in this
chapter, in that it demands an ability to recognise and coordinate the
contributions of two discrete variables to the behaviour of the beam.
If the tasks were more heterogeneous than in much previous work, so
too was the age range. The children involved in these studies ranged from
seven to thirteen years of age, but there was little indication that this in
itself was a signiWcant factor in shaping the results. The tight tie between
the mechanism of peer facilitation and a particular phase of intellectual
development, suggested by the socio-cognitive conXict theory, sits rather
uneasily with our results. It seems rather that whatever processes are
involved operate in a rather similar fashion across a considerable range
both of tasks and age levels.
Broadly, the Wndings of these studies do seem to conWrm that the
pairing of children while they work on a problem can improve their
performance on similar problems when they meet them on their own at a
later date. The Wrst Tower of Hanoi study showed that such facilitation
can occur even for the initially more able member of a pair of children,
but is likely to occur only for children who from the outset have at least
some ideas of their own about how to tackle the problem. The second
Tower of Hanoi study indicated that the beneWts of peer interaction
depended on keeping the children jointly engaged and preventing domi-
nance and ‘turn-taking’. The third Tower of Hanoi study conWrmed this
Wnding in relation to a computer-based version of the task, though the
gains here were limited to the speciWc trials used in the paired session.
The Tower of Hanoi studies were characterised by very low levels of
task-related talk between the members even of the most eVective pairs. By
contrast, Mastermind elicited high levels of discussion, and, in some
pairings, high levels of disagreement about what entries to make and why.
The level of such argument turned out to be associated quite clearly with
the pre- to post-test gains made by the members of such pairs. Similarly,
the balance beam task lent itself to verbal justiWcation of the proposed
26 Social processes in children’s learning
solutions, and the two studies reported here both suggest that it was the
combination of the availability of a partner and the requirement for such
justiWcation which resulted in the highest levels of progress. The fact that
the verbal articulation of a justiWcation was not in itself suYcient to
produce such progress suggests that it is the process of agreeing a justiW-
cation that holds the key to the productivity of this condition. Reconciling
disagreements might of course be a part of this process.
Thus, the Wndings reported here are consistent with the idea of produc-
tive conXict of views between partners, but the conception of conXict
required is rather broader than Doise’s original socio-cognitive conXict
hypothesis seemed to suggest. The presence of diVering points of view
may be a necessary ingredient of productive conXict, in this sense, but so
too may be a means of securing and sustaining the joint engagement of all
the parties in the resolution of disagreements. The focus on verbal inter-
action and on the co-construction of solutions to tasks, more characteris-
tic of Vygotskian approaches, also Wnds some support in these studies.
Of course, the series of studies revisited here represent only one of
many possible approaches to the empirical study of peer interaction of
children’s learning, and are subject to many limitations. The studies were
conducted in school, but not in the context of ongoing classroom activity,
nor by the regular classroom teacher. For the most part they involved
random allocation of children to experimental conditions, thus riding
roughshod over such considerations as friendship patterns. The Wnal
study suggested that this was not an irrelevant consideration.
Most of the studies described also focused not on pair or group per-
formance as such, but on individual pre- to post-test performance
changes. These were used as a measure of facilitation of individual
learning as a result of typically single and rather brief periods of paired
interaction around a well-deWned task.
We shall trace the further development of this line of research work in
Chapter 4, but meanwhile in the following chapter we shall turn to look at
a speciWc context for learning, namely learning with computers. The
studies described above already reXect a movement towards task presen-
tation using computers, which has a variety of practical advantages for the
researcher. By the late 1980s, it became apparent that computers were
going to Wgure prominently in the learning experiences of both children
and adults, and researchers began to show considerable interest in the
special characteristics and possibilities of this kind of learning. This
interest has shaped the direction of subsequent research on social pro-
cesses in learning to a considerable degree, so a brief review of the
‘computers and learning’ literature seems appropriate at this juncture.
3 Computers and learning
Introduction
A comprehensive review of the research literature on what computers
might have to oVer for learning would have to go back a long way to Wnd
its starting points (e.g. Suppes, 1966; see also Light, 1997). However, at
least in the in the UK, it is only relatively recently that children have been
able to secure reliable access to at least one classroom computer, most
teachers have received some training in the use of computers, and the
school curriculum envisages computer use in most subject areas (Crook,
1994). The prominence given to computers, and the substantial invest-
ment of resources entailed in their use, are widespread features of educa-
tion across the developed world. The psychological theories of learning
which have informed the development of educational computer use over
the last thirty years or so oVer a fair reXection of the psychology of
learning more generally across this period. In this chapter we shall explore
some of the ways in which psychological accounts of children’s learning
have contributed to, and been reXected in, the way computer-based
learning has developed.
and simulations. Schank and Cleary (1995), for example, draw an anal-
ogy with Xight simulators, which assist pilots to learn to Xy by providing
realistic environments without all the risks and costs of the real thing. In
much the same way, they argue, other types of experience can be
simulated on the computer, allowing children to play, experiment and
explore, and greatly extending the range of things children can learn
by doing.
Multimedia computer simulations can allow children to explore the
inside of the human body, or the outer reaches of the universe. ‘Alterna-
tive’ worlds can be simulated, in which, for example, diVerent laws of
motion apply (Smith, 1991). Developments in the technology of virtual
reality may extend the possibilities of such ‘simulated worlds’ very rapidly
in the future. Equally, other developments in the Weld of hypermedia are
already bringing to the classroom a much wider range of learning re-
sources, along with an educational philosophy which stresses the need to
‘put the learner in control’ (Hutchins, Hall and Colbourn, 1993).
Introduction
In this chapter we shall describe a series of experimental studies based on
the collaborative use of computer-based planning tasks. The studies are
linked by the particular type of task used. Rather than start from existing
tasks, as in the Wrst series of studies (Chapter 2), we decided to develop
our own. Instead of relying on computer simulations of non-computer
tasks, such as the Tower of Hanoi or the balance beam, we wanted to
develop a task which would better exploit the potential of the computer.
We also wanted a task which would allow a more extended experience of
solo or collaborative problem solving, rather than a sequence of quick
trials.
We were interested by evidence from RogoV and colleagues (Gauvain
and RogoV, 1989; Radziszewska and RogoV, 1988) that the ‘metacogni-
tive’ processes involved in successful planning (e.g. conscious, explicit
monitoring and careful control of plan development) might lend them-
selves particularly well to facilitation through interaction, discussion and
collaboration. We were also aware of evidence that ‘adventure games’
seemed to oVer particularly stimulating contexts for peer discussion and
interaction (Crook, 1987; Johnson and Johnson, 1986).
The task we designed lies somewhere along a continuum running from
genuinely open adventure games to closed logical problems of the Tower
of Hanoi type that we had used previously. We wanted a well-structured
problem, but one which at the same time oVered a variety of routes to
solution, some more eYcient than others. We were also seeking to cap-
ture several important aspects common to many situations of learning
with computers. For example, computers may often hold a large amount
of information which needs to be located and used systematically as a
basis for completing the given task. The information is used as a basis for
formulating a plan, which may then need to be amended in the light of
feedback in the course of doing the task. We felt that this combination of
32
Kings, Crowns and Honeybears 33
Figure 4.2 King and Crown software: the general information screen
Kings, Crowns and Honeybears 35
Figure 4.3 King and Crown software: the pilot information screen
Figure 4.4. They are then required to specify the point of departure, a
destination, the characters being moved (and whether or not they are
moving the crown) and the means of transport being used. The move is
then made by clicking on ‘Go’.
The task is made complicated by the presence of bandits and pirates.
The bandits will steal the crown on the mainland unless the guard is
present. The pirates will steal the crown from any ship sailing the sea even
if the guard is present. If the users obtained this information at the outset
and thought it through, they could correctly conclude that the crown
must be retrieved by plane. However, the plane can only carry a pilot and
one passenger, which complicates things further. The optimal solution, in
fact, is to take the car (and driver, guard, captain and pilot) to Crowmar-
ket; to take the ship (and captain and pilot) to Hushley; to take the plane
(and the pilot and the captain) to Fruggle to collect the crown; to return
to Crowmarket, and then for all to return by car to Ashlan. This repre-
sents a total of Wve moves.
The software (initially developed by Richard Joiner) automatically
updates all relevant information as each move is made. Thus at any
point the children can stop and take stock of where the characters and
means of transport are. Whilst in theory this might not be necessary, in
practice it usually is. Most users, at least initially, either set oV in the
wrong direction or take the wrong characters, and thus encounter diY-
culties. If the users attempt to make a move which is impossible, or
which would lead to the pirates stealing the crown, they get a warning
message and are prevented from actually making the move. This means
that the users then have to replan the move, to take account of the
particular problem they have encountered. The task is thus a diYcult
one. Nevertheless most users, whether children or adults, Wnd it engag-
ing and highly motivating.
In an initial study using the ‘King and Crown’ software we set out to
investigate whether children of around eleven years of age would proWt
from working collaboratively on this task (Blaye, Light, Joiner and Shel-
don, 1991). First we wanted to know whether, on a task such as this
which makes heavy demands on information management and planning
skills, those working in pairs would perform better than those working
alone. Second, we wanted to see to what extent any advantage shown
by the pairs would transfer to a subsequent assessment of individual
performance.
The study involved thirty-nine eleven-year-old children. They were
taken from their school classroom to another room in the school either
in pairs or individually (thirteen pairs and thirteen individuals). The
decision as to which should work alone and which in pairs was made
Kings, Crowns and Honeybears 37
Close scrutiny of the strategies adopted suggests that very few of the
children even in this second session actually planned out a successful
strategy in advance. In fact ten of the pairs and eleven of the individuals
tried to move the crown by ship, but were stopped in their tracks by the
warning message about the pirates. The diVerence between pairs and
individuals was that, following this setback, six of the pairs but only two of
the individuals immediately turned their attention to the plane. Thus it
seems that the advantage of the pairs may lie more in adaptive re-planning
than in thorough pre-planning.
The videotapes were analysed in terms of who, within the pairs, con-
trolled the mouse most, and how decisions were made. Mouse control in
the paired sessions (sessions one and two) turned out not to be correlated
with score on the individual post-test (session three), so it did not seem
that this aspect of the interaction was strongly related to ability or learn-
ing. However, the lack of an individual pre-test meant we were unable to
rule out the possibility that improvement was related to the extent of
mouse control.
Decisions regarding the destination for the next move, or who was to be
taken, or what means of transport was to be used, were examined for all
pairs in the second session. Intercoder reliability was checked on a ran-
dom sample of six pairs. A particular child was identiWed as the author of
a particular decision if he or she acted without reference to the partner or
suggested a choice which the partner acted on. Intercoder reliability was a
little over 80 per cent. Each of the children involved was thus given a
‘decision score’ based on the proportion of all decisions that child was
responsible for. Decision score did turn out to be signiWcantly correlated
(−0.45) with individual performance on the third session. The children
who controlled more of the decisions in the second session tended to do
better as individuals on the third. Decision score was not related to any
tendency to control the mouse more when working with a partner.
Decisions which were formulated simultaneously, or where both part-
ners contributed some part of the formulation, were coded as shared
decisions. It is interesting that shared decisions were signiWcantly more
prevalent in pairs in which both partners subsequently succeeded on their
own. Pairs can also be scored in terms of the extent of disparity in the
decision scores of their members. If we look back to the crucial point in
the task alluded to earlier, where pairs were confronted with the impossi-
bility of taking the crown by ship, it turns out that the six pairs who
‘adaptively replanned’ to take the plane at that point were the six with the
lowest decision disparities. This suggests that it is not just having a
partner that makes the diVerence, but having a fairly symmetrical rela-
tionship with that partner as far as decision making is concerned.
Kings, Crowns and Honeybears 39
Figure 4.6 The mean level of performance for the ‘pairs’ and ‘individ-
uals’ during each of the three sessions
pairs). In others there was a diVerence of more than one level in the
pre-test performance of the pair members (‘highly asymmetrical’ pairs).
As it turned out, the pairs established at stage two actually comprised
fourteen symmetrical pairs, twenty-three asymmetrical pairs and eight
highly asymmetrical pairs. We analysed the progress made (from stage
one to stage three) for members of each of these types of pair. The average
improvement in performance level from stage one to stage three averaged
3.8 for members of symmetrical pairs, 2.9 for members of asymmetrical
pairs and 2.4 for members of highly asymmetrical pairs. This diVerence
proved to be statistically signiWcant overall. The outcome for the symmet-
rical pairs was signiWcantly better than for either of the other two groups.
As a check on this rather striking result, we went on to examine the
performance of a sub-sample of children who all started from the same
pre-test level. The modal pre-test score (level one) was used for this
purpose. Thus, one group consisted of those children whose pre-test
score was one and who were paired with a child whose pre-test score was
less than their own (n = 7). Another group consisted of those whose
pre-test score was one and whose partners’ score was greater than their
own (n = 16). The third group were those whose partner also scored one
at pre-test (n = 12).
Figure 4.7 shows the results of a comparison between these three
groups. The best stage two performance came from those children work-
Kings, Crowns and Honeybears 47
this study. One hypothesis might be that the lack of stage three (post-test)
advantage for those who worked in pairs might have arisen because, by
contrast with the Wrst King and Crown study, the children in the present
study all worked alone at the Wrst (pre-test) stage. Thus the degree of
diVerentiation between the paired and individual conditions was less in
this study. It is also possible that the fact that children meet the task Wrst
on their own, and only later in pairs, alters the dynamics of the children’s
interaction with the task.
However, there is another design diVerence which oVers a possible
explanation for the diVerence in outcome, and one which lends itself
more readily to testing. This concerns the way in which the ‘individual’
sessions were handled. As noted earlier, whereas the children working
individually in the King and Crown studies were alone with an adult
experimenter and a computer, those in this study were always in a room
with three others classmates working on the same task. There was no
opportunity for overt interaction between the children during the session,
but might this simple fact of peer presence have made a diVerence? There
is some suggestive evidence on this point from a study of secondary
school physics students using a computer simulation (Whitelock, et al.,
1993). These authors report a facilitation eVect when students worked in
the presence of others, even without interaction, as compared with work-
ing entirely on their own. The next study was designed as a direct check
on whether this kind of facilitation was occurring here.
The results of this study were striking. The mean performance level for
the children who worked in the presence of others (2.9) was signiWcantly
better than the performance of those who worked alone (1.8). Thus quite
apart from any consequences of productive verbal or practical interaction
between partners working together on the task, we have here evidence for
the operation of a quite diVerent kind of social facilitation of perform-
ance, which depends simply on the presence of classmates working on the
same task.
Overview
The three studies using the King and Crown software gave a more
encouraging picture of the beneWts of peer interaction than the studies
using standard problem-solving tasks reviewed in Chapter 2. With this
more open and extended type of problem solving (arguably more ‘face
valid’ in relation to educational experiences), the present studies seem to
show that both children and adults can reap considerable beneWt from the
presence of a collaborating partner. This peer advantage was evident even
though the children were given no more than a general invitation to work
collaboratively, without being forced into any particular pattern of inter-
action through the instructions or the structuring of the task itself.
The Wrst King and Crown study showed that the peer advantage could
carry over to individual subsequent performance on a slight variant of the
task. The second showed that the verbal interactions within the pair were
predictive of pair performance, of the subsequent performance of the
individual members of the pair, and indeed of the relative performance of
members of the pair at post-test. The third attested to the age indepen-
dence of at least some of the eVects observed, and suggested that the more
eVective use of available information to build a shared representation of
the task might be one factor explaining the advantage enjoyed by pairs.
The Wrst Honeybears study was to have been the deWnitive one. Its
scale, and full pre-test/post-test design, should have allowed a detailed
analysis of the basis of paired advantage. The observation that symmetri-
cal pairings were more eVective than other pairings seems to support an
interpretation couched broadly in terms of coconstruction rather than
conXict. However, other studies (e.g. Howe and Tolmie, 1999, using
tasks which focus on the judgements and prior knowledge that children
bring to physics tasks) have found evidence that asymmetrical pairs are
more eVective. We should acknowledge that diVerent psychological and
interactive processes are likely to be brought into play by diVerent types of
task, and we should not seek to draw overgeneralised conclusions. The
more important observation is that, as a whole, pairs did produce better
Kings, Crowns and Honeybears 51
puters have been noted in various studies. Thus, for example, it has been
argued that while boys tend to adopt a more analytic and closed approach
in working with computers, girls tend to adopt a more open-ended and
exploratory approach (Turkle and Papert, 1990). Kirkup (1992) has
suggested that, whilst either style is compatible with using computers for
learning, software development for learning as well as for leisure has
tended to follow a masculine path.
Another stylistic diVerence sometimes reported points to a natural
linkage between the issue of sex diVerences in response to computers and
the issue of peer interaction in learning. Girls, it has been suggested,
prefer collaborative modes of working, while boys prefer to work alone
(Dalton, 1990; Hoyles, 1988). Hoyles, Sutherland and Healy (1991)
suggest that while boys seem to see interactions around the keyboard as
time-consuming diversions, girls see them as aVording opportunities for
mutual support and the development of ideas. Observations in mixed sex
classrooms suggest that, given the opportunity, boys tend to work individ-
ually at the computer, while girls tend to work cooperatively (Underwood
and Underwood, 1999). When children do work in groups around the
computer, girls tend to work more collaboratively than boys. The boys
tend to get fractious, one result of which is that they get more of the
teacher’s attention than the girls do (Culley, 1993).
Mixing the sexes does not seem to help. A substantial literature points
to the danger that boys will dominate mixed-sex interactions (Graddol
and Swann, 1989; Swann, 1997). The research literature more speciW-
cally on mixed-sex groups (mainly pairs) working with computers sug-
gests that these may indeed be relatively unproductive, with boys being
socially dominant in such situations (Siann, Durndell, McLeod and
Glissov, 1988). Watson refers to classroom observations such as: ‘Boys
came in Wrst and sat squarely in front of the screen. I had to remind them
to let girls have room to sit down’ (1997, p. 221).
Underwood, McCaVery and Underwood (1990) found that while
single-sex pairs did better than individuals when working on a sentence
completion task , the same was not true for mixed-sex pairs. Underwood,
Jindal and Underwood (1994) found that mixed-sex pairs of upper pri-
mary age children performed poorly on a computer-based language task,
showing little evidence of cooperative working, even given encouragement
to cooperate. Howe and Tolmie (1999) found with twelve to Wfteen year
olds working on a computer-based physics task that the amount of
task-related talk (expressed as number of dialogue turns) was substantially
lower in mixed-sex as compared to single-sex pairs. They also observed
that such interaction as there was in the mixed-sex pairs was not related to
learning outcomes in the way that interaction in single-sex pairs was.
Gender agendas 55
badly as those in the girl–girl pairs. We thus seem to see a substantial main
eVect for gender upon performance, but no eVect of mixed versus single-
sex pairing.
One of the post-experimental questionnaire items asked the children to
indicate on a Wve-point scale how much they agreed or disagreed with the
statement: ‘I like working on my own more than with a partner’. Regard-
less of pair type, the boys tended to agree with the statement, the girls
to disagree.
Seating positions were interesting. No attempt was made to regulate
this, but it transpired that in the mixed-sex pairs the boys were signiWcant-
ly more likely than the girls to end up sitting in a position (on the left)
which gave them right-handed access to the mouse. Whether this is
because the boys sought out this chair or because the girls avoided it we
are unable to say.
The children were given a general injunction to share the mouse, but
the evenness of ‘shares’ was variable. The girl–girl pairs showed the most
even distribution, the boy–boy pairs were intermediate, and the mixed-
sex pairs showed the least even distribution. The median number of
switches of mouse control between the members of girl–girl pairs in
the course of the twenty-Wve minute session was eleven. In the boy–boy
pairs the corresponding Wgure was four, and in the mixed pairs the
median was zero. In most of these pairs the boy controlled the mouse
throughout.
Thus there were quite marked sex diVerences in performance, and
quite marked pair-type diVerences in some non-verbal aspects of behav-
iour. However, our attempts to diVerentiate the pairs using the categories
of verbal exchange described in the previous chapter were not successful.
For example, the greater success of the boys was not associated with
signiWcantly higher levels of verbally explicit planning or negotiation.
These Wndings led us to consider the extent to which the noticeably
masculine characteristics of the King and Crown software might be
responsible for the gender-related diVerences in the apparent learning
outcomes. On the post-experimental questionnaire there was no indica-
tion that the girls were less attracted to the task, but this merely reXected
a ceiling eVect, in that all children tended to give very high ratings.
Objectively, the task consisted of an adventure game in the form of a
‘quest’ in which male characters engaged in stereotypically male activ-
ities. What would happen, we wondered, if we re-versioned the task so as
to avoid these features? The result was the Honeybears version of the
task, described in the last chapter. In attempting to produce a gender-
neutral version of the task we used images from fairy tales and advertise-
ments. Some commentators (e.g. Fitzpatrick, 1996) see the resulting
Gender agendas 59
Figure 5.3 The mean level of performance for the three diVerent pair
types
Wfty-two children, aged eleven, half boys and half girls. Half the boys and
half the girls were randomly allocated in advance to each version of the
task. The children were withdrawn from the class in groups of four (two
boys and two girls) and they went to a room in which there were four
computers. The two boys sat at one machine to start with, and were
introduced to one version of the task. They then moved to separate
machines and worked on that version of the task on their own. The two
girls, meanwhile, were introduced to the other version of the task by a
second adult, and went on to work on this version on their own. One of
the adult experimenters was male, the other female, and they balanced
their roles with respect both to the gender of the children and the version
of the software they were working with.
In the initial brieWng the children were given a tightly scripted tutorial
introduction in which they were shown the goal of the task. They were
also shown how to retrieve the information available in the software, how
to make a move, and the consequences of attempting to make an imposs-
ible move. Finally, the children were informed of the time available (thirty
minutes), but told not to worry if they did not solve the problem.
Figure 5.4 shows the average scores for the two gender groupings and
two versions of the task. Overall, performance was signiWcantly better on
the Honeybears than on the King and Crown version, but there was no
signiWcant eVect of gender. However, while the performance of the boys
remained virtually unaVected by the software type, the performance of
the girls was far superior when using the Honeybears software, resulting
in a signiWcant interaction eVect. Indeed, the girls’ mean level of perform-
ance on Honeybears slightly exceeded that of the boys, although this
diVerence is not statistically signiWcant.
The results clearly suggest that the performance of the girls was strong-
ly inXuenced by the version of the task employed. However, the study
suVered two weaknesses. Since the Honeybears version of the software
was not originally designed to make a comparison with the King and
Crown, it diVered in a number of ways, for example in the layout of the
various screens and the wording of the messages. These diVerences
potentially confound the present comparison. The facilities for automatic
computer recording of the children’s actions also diVered, precluding
detailed comparison of the task solution strategies adopted by the
children.
To deal with these problems, as well as to see how replicable the
previous result would prove, we designed a new version of the King and
Crown task called Pirates. This kept the storyline of the King and Crown
task, but it shared all design and interface characteristics with Honey-
bears. Using Pirates and Honeybears as the contrasting versions this time,
62 Social processes in children’s learning
Figure 5.4 The mean level of performance for the girls and boys on the
King and Crown and Honeybears versions of the task
we took another sample of eleven year olds (forty-eight this time) and
re-ran the study just as before. Half the boys and half the girls were
randomly allocated in advance to Pirates and half to Honeybears.
The results are shown in Figure 5.5. This time there was no overall
eVect of software version, nor was there any overall eVect of gender. The
interaction between the two, however, remained signiWcant. As before,
the boys were apparently little aVected by which version of the task they
encountered, whereas the girls did far better with the Honeybears ver-
sion. On average, the girls outperformed the boys on Honeybears, though
this diVerence was not statistically signiWcant.
Inspection of the computer traces showed no signiWcant eVect of
gender or software type, nor any interaction between these, in terms of
how long it took the children to get started on the task, how much
information they searched before beginning to make moves, or in the
direction of initial moves. The only gender by software interaction dis-
covered from the trace concerned errors associated with the characters in
the scenario. Such errors might arise, for example, when a child failed to
Gender agendas 63
Figure 5.5 The mean level of performance for the girls and boys on the
Pirates and Honeybears versions of the task
More generally, the evidence that mixed-sex pairs and single-sex pairs are
marked by diVerent styles of interaction is much clearer than the evidence
that children in mixed-sex pairs perform less well or learn less.
Fitzpatrick (1996), for example, showed in a number of studies that
interactional patterns in mixed-sex pairs were characterised by con-
strained interaction, greater task demarcation and less collaborative talk.
However, the girls in these pairs showed no less progress in understand-
ing at post-test than girls who had worked in single-sex pairs. Such results
should caution us to recognise that the relationship between interaction
and learning is far from straightforward.
We saw in the previous chapter (Wrst and second Honeybears studies;
Light et al., 1994) that children tended to perform better on the Honey-
bears task when another child was present in the room working on a
similar task than when they were alone with the experimenter. In those
studies the children present were always of the same gender. We went
on to consider whether mixed-sex pairing might make a diVerence even
here, where there was not overt interaction (we called this a ‘coaction’
condition).
Sixty-two eleven-year-old children participated in our Wrst study of this
issue (Light, Littleton, Bale, Messer and Joiner, in press). The children
were assigned at random to ten boy–boy pairs, ten girl–girl pairs and
eleven mixed-sex pairs. They were taken in their assigned pairs from their
classroom to another room in the school by an unfamiliar (male) experi-
menter, who gave a brief introduction and demonstration of the Honey-
bears task and then sat each of them at a computer. The children were
told that they probably would not Wnish the task before the end of the
session, but that they should see how far they could get.
The two computers were close to one another, but arranged so that the
children could not see one another’s screens. They could see one an-
other’s faces, but were asked not to talk to one another during the twenty
minutes they spent working on the task. The adult remained present
throughout the session, but did not intervene in any way. The study
involved only a single session for each child.
Figure 5.6 shows the pattern of performance. Boys did slightly better
than girls overall, but more interestingly there was a signiWcant gender by
pair type interaction. Boys tended to do better when their ‘partner’ was a
girl, but girls tended to do less well when their ‘partner’ was a boy. The
trace data reveal similar statistical interactions in respect of information
searching. Girls in mixed-sex pairs spend less time searching for informa-
tion (both before the Wrst move and overall) than girls in single-sex pairs,
whereas the boys spent more time in information searching than boys in
single-sex pairs. Thus it seems that in the mixed-sex pairings, the boys are
66 Social processes in children’s learning
Figure 5.6 The mean level of performance for the mixed and single-sex
dyads
behaving more planfully, but the girls less planfully than in single-sex
pairings.
The surprising feature of these results is that, whereas in previous
studies children in mixed-sex pairs actually performed at very much the
same level as those in single-sex pairs, here we are seeing a signiWcant
eVect of pair type on performance. But in this case, unlike those which
have gone before, the children are not working collaboratively at all, but
simply working in proximity with one another. Thus despite evidence
that in interactive mixed-sex pairs the boys do indeed tend to dominate
the interaction with the computer, we Wnd clearer evidence for the eVect
of partner gender on performance when partners do not interact around
the task than when they do.
On a post-experiment questionnaire, the children were asked whether
they thought boys or girls were most interested in, and better at using,
computers, or whether they thought both were equal. About half of the
children (boys and girls equally) thought that boys were more interested
in computers than girls, whereas no child thought the opposite. Only
about one third of the children (boys and girls equally) thought that boys
were better at using computers than girls, but again no child thought the
opposite. Thus we can say that between a third and a half of the children
Gender agendas 67
Figure 5.7 The mean level of performance for the mixed and single-sex
pairs for the interacting pairs
Figure 5.8 The mean level of performance for the mixed and single-sex
pairs for the coacting pairs
Overview
In the introduction to this chapter we provided a brief review of the now
quite extensive research literature focusing on gender diVerence in re-
sponse to computers. Such gender diVerences are all seen as working to
the disadvantage of girls. The pattern of gender diVerence observed with
our King and Crown task seemed to conWrm the gloomy picture; the boys
dominated the mixed-sex interactions and performed signiWcantly better
than the girls on the task at post-test. Nonetheless, it is important to note
that the extent of the gender diVerence in performance was independent
of whether the children had worked in same-sex or mixed-sex pairs.
The introduction of the Honeybears version of the task was marked by
a striking reduction in both the gender diVerence in performance and the
dominance of the computer interface by the boys. Controlled contrasts
of the two versions of the task showed quite clearly that, even with
exactly the same underlying task, the setting of the task made a great
deal of diVerence to the performance of the girls. The gender diVerence
which was so striking in the King and Crown task could be ameliorated
or indeed eliminated by simply altering the storyline and characters
involved. Whether versions of the task could be found which create
Gender agendas 71
raised the level of general drive, and favoured the production of the most
readily available response to the task in hand. With simple tasks, this
response is likely to be correct, so that performance facilitation results.
With more complex tasks, successful solution may depend on the inhibi-
tion of such dominant responses, and raised levels of general drive may
thus reduce the quality of performance.
Cottrell, Wack, Sekerak and Rittle (1968) argued that it was not so
much the mere presence of individuals that makes a diVerence, but rather
the ‘evaluation potential’ of the audience. They showed, for example, that
for individuals performing a task, the presence of others made no diVer-
ence if they were blindfolded, but it did make a diVerence if they were
watching the performance keenly.
Both the pattern of Wndings reported by Zajonc and his approach to
explaining them have found mixed support in subsequent research (Bond
and Titus, 1983; Glaser, 1982). The predicted pattern of results for
simple and complex tasks does not always occur, especially in relation to
tasks involving memory and cognition (Hartwick and Nagao, 1990).
There have been a number of attempts to reconceptualise the issue of
social facilitation in more cognitive terms. For example, Manstead and
Semin (1980) argued that the presence of the other could be seen as
making demands on Wnite attentional resources, to the detriment of
learning tasks calling for considerable attention. ‘Automated’ performan-
ces, on the other hand, demand little attention and may beneWt from
anything which maintains alertness.
This focus on attentional resources and their disposition underpins the
approach of Monteil and Huguet (in press), who report an unpublished
study in which Huguet, Monteil and Galvaing used the Stroop test to try
to distinguish dominant response explanations from explanations based
on attention. With, for example, the word (e.g. ‘red’) written in a diVerent
colour (e.g. green), the participant is asked to name the colour. The
correct answer is thus ‘green’, but the interfering dominant response (to
read the word) will tend to generate the erroneous response ‘red’.
In these circumstances, if the presence of another person increases the
tendency to generate the dominant response it should lead to a decrement
in performance. However, if the presence of another person makes it
harder to focus attention, as Monteil and Huguet suppose, then it might
be expected to improve performance. The results showed that provided
the other person was attentive and looking at the participant, Stroop
interference was reduced and performance improved.
But Monteil and Huguet are seeking to foster a position which is not
only more cognitively oriented than Zajonc’s, but also more socially
oriented. They emphasise the importance of the participant’s personal
76 Social processes in children’s learning
perspective on the task and the broader situation within which it is set.
Participants come to learning situations with a history of experience of
their performance across a range of more or less related tasks in more or
less related situations; an ‘autobiography’ of themselves as learners.
The central point here is that the presence of another can generate
diVerent signiWcations according to the history of the individual learner.
Social contexts do not exist independently of the individual. Rather:
‘Social contexts exist only through the intervention of cognitive structures
of contextualisation, such as those linked to the autobiographical memory
of the individual’ (Monteil and Huguet, in press). In other words, while
individual performance depends upon social contexts, social contexts can
only be understood in terms of individuals.
Monteil and Huguet (in press) point out that a number of studies over
the last twenty years or so have suggested that the experience of the
individual immediately prior to the peer presence situation can make a
signiWcant diVerence to the outcome observed. For example, Mash and
Hedley (1975) showed that a positive interchange with the experimenter
led to facilitative eVects of subsequent peer presence on a simple motor
task, but a negative interchange with the experimenter resulted in inhibi-
tory eVects of subsequent peer presence. Similarly, Seta and Hassan
(1980) showed that audience eVects could vary according to the individ-
ual’s immediately prior history of success and failure on the task. Using a
memory task, they showed that the presence of an audience improved the
performance of those with a previous history of success but impaired that
of those who had encountered failure.
If social comparison is one key concept in twentieth century social
psychology, ‘social loaWng’ is another. Ringelmann (1913) observed that
the total physical eVort produced by a group (in response to a task
involving pulling on a rope) was less than the sum of the eVort put in if
each member of the group worked alone. The diVerence was found to be
proportional to the size of the group. The term ‘social loaWng’ was
subsequently introduced to refer to the tendency of individuals to pro-
duce less eVort in group than in individual situations. It has been shown
that social loaWng does not occur in a situation where the individual
contributions of the participants are expected to be identiWable (Will-
iams, Harkins and Latane, 1981). It seems that evaluation (or the poten-
tial for evaluation) of relative performance is a determining factor in this
regard.
Monteil and Huguet (in press) are critical of research in this area
because of its characteristic neglect of the history of experience of the
individual participant. They cite a study by Sanna (1992) in which
individuals were paired for a vigilance task. Having received feedback
Social comparison and learning 77
indicating that they were doing better or worse than their partner, they
went on either to work as a pair (where their success would be judged on
‘group product’) or coactively (where their individual success levels
would be visible). Those who had been favoured by the initial compari-
son subsequently performed better in the coaction than the group condi-
tion (a result consistent with social loaWng). However, the reverse was
true of those whose initial feedback had suggested that they were doing
less well.
It seems, then, that social comparisons can moderate the eVects at-
tributed to social loaWng. So too can individual diVerences of a more
general/stable kind. Huguet, Charbonnier and Monteil (1995) showed
that reduced productivity in a collective as against a coactive situation was
characteristic only of students who were high on ‘self-uniqueness’, as
indexed by a separately administered questionnaire. Extending this to
cultural diVerences, Gabrenya, Wang and Latane (1985) have shown in a
comparison of Chinese and North American students that social loaWng
was characteristic only of North American males. Chinese males showed
exactly the opposite pattern (‘social striving’). Without wanting here to
open up the complex debate about individualistic and collectivist cul-
tures, we can simply recognise that universal accounts which take no
account of individual, social and cultural characteristics are unlikely to be
adequate to the phenomena. For example, social loaWng appears to be
more a male than a female phenomenon: Karau and Williams, 1993, and
Monteil and Huguet, in press, report data indicating that while boys
produce better memory recall performance in ‘public’ than in anonymous
conditions, the reverse is true for girls.
condition, but after the lesson all students completed a written test on the
material taught.
In the ‘No social comparison’ condition, the high attainers did better
on the test than the low attainers, and the individuation/anonymity
manipulation had no eVect. In the ‘Social comparison’ condition, how-
ever, individuation led to an ampliWcation of the diVerence between the
high and low attainers, while anonymity reduced the diVerence to almost
nothing. Given the combination of social comparison and anonymity, the
high attaining pupils did rather less well than they had done in other
situations, while the low attainers did exceptionally well.
Similar results were obtained in a second study conducted with only
high attaining pupils. Here, social comparisons were introduced by giving
pupils feedback on a prior task, indicating success or failure. Even when
this feedback was actually uncorrelated with real performance, it engen-
dered the same pattern of results. Negative feedback was followed by
better learning under anonymous than under individuated conditions,
while positive feedback was followed by better performance under indi-
viduated than under anonymous conditions (Monteil, 1988).
Subsequent studies showed that the eVects shown in these initial
studies are also conditioned by the academic context in which the task is
set, being evident for highly valued disciplines such as mathematics and
biology, but not for lower valued disciplines such as ‘technical and
manual education’. This observation opened up a new line of research, in
which Monteil and Huguet (1991) explored the eVects of representing a
given standard task to some students as ‘geometry’ and to others as
‘drawing’. The actual task was always the same and involved a complex
rectilinear Wgure (see Figure. 6.1), which the pupils were shown for a
short period of time and asked to remember. Their success on both a
recall (which involved drawing from memory) and a recognition task was
shown to vary as a function of both discipline context and their own
academic standing. When the task was represented as being ‘geometry’,
there was a large diVerence in memory performance between the children
classed as generally high attainers and those classed as low attainers,
with the high attainers doing better. However, when the task was repre-
sented as ‘drawing’ there was no diVerence in performance between the
two groups.
This intriguing result prompted us to attempt a replication in the UK.
It seemed possible that the results reported by Monteil and Huguet might
depend upon certain features of the French education system which
diVerentiated it from the British system. In particular, it appears that
pupils are made more explicitly aware of their own levels of academic
performance in France than in the UK. For example, they are required to
Social comparison and learning 79
For the high attainers, the modal ranking of importance for maths was
1 (i.e. it was seen as the most important subject of the nine subjects
listed). For the low attainers the modal ranking of the importance of
maths was signiWcantly lower, at sixth in the list of nine. However, both
high and low attainers put art lower down the importance ranking than
maths, with a modal ranking of 8 from both groups. For the high at-
tainers, maths was seen as signiWcantly more important than art, whereas
for the low attainers this diVerence was not signiWcant.
Overall, the higher the relative importance children attached to the
subject in which they were supposedly being tested, the longer they
tended to spend on the reproduction of the Wgure from memory. This
relationship did not quite reach a satisfactory level of statistical signiW-
cance, but again seems to indicate that taking a longer time is indicative of
greater eVort and attention.
When asked to rank the subjects in terms of how good they thought
they were at them, the high attainers gave themselves a modal rank of 2
for maths, whereas the lower attainers gave themselves a modal rank of 9
for maths, indicating that this was the subject they were least good at.
However, there was a great deal of individual variability, and this diVer-
ence was not reliable. For art, however, there was a statistically reliable
diVerence between the rankings the low attainers gave themselves (modal
rank 1) and those the high attainers gave themselves (modal rank 9).
Taken together with Monteil and Huguet’s earlier Wndings, these
results testify to the extent to which the apparent disciplinary context of a
task can make a diVerence even when the task itself remains exactly the
same. The success children have with academic tasks clearly can depend
upon how the task is labelled and how the children perceive their own
abilities in relation to that label. More generally, the interaction between
levels of attainment and disciplinary context shows once again that given
contexts depend for their eVects upon individual attributes and histories.
In an extension of this work with the memory task, we returned to the
question of learning with computers, and to the issue of gender. The
Wgure shown in Figure 6.1 was presented to sixty-four eleven year olds,
half of them male, half female. To half of the boys and half of the girls, the
Wgure was presented as before, on paper. To the other half it was pres-
ented on a computer screen. In all cases, the children were required
simply to look at it, and then to reproduce it from memory on paper. Thus
even in the ‘computer’ condition, the children did not actually have to do
anything with the computer at all.
Children came to the testing room in same-sex pairs, but worked
separately, without interaction and without being able to see what each
other were doing. Either both were in the ‘computer’ condition, or
82 Social processes in children’s learning
Figure 6.2 The mean recall score for the girls and boys for the two
modes of task presentation
neither were. If they were in the ‘computer’ condition, the task was
prefaced by some general questions about whether they enjoyed working
with computers, whether they had one at home, and so on.
The pattern of results in terms of recall scores is shown in Figure 6.2.
As can be seen, the girls did better when the task was presented on paper
than when it was presented on computer, while the reverse was true of the
boys. This resulted in a statistically signiWcant interaction between gender
and condition eVects in an analysis of variance. A simple main eVect
analysis showed the gender diVerence in the ‘on paper’ condition to be
statistically signiWcant.
A similar pattern of results was shown in terms of a diVerent test of
memory, which involved recognition. This was done after the recall task,
and involved pupils being presented with a series of twenty paired images.
Members of each pair were identical except in orientation. Children had
to decide whether the image represented a fragment of the original Wgure
or not, and if so, which image was correctly oriented.
For children in the ‘on paper’ condition, the paired images were
presented as a booklet, with a pair of images on each page. For children
Social comparison and learning 83
Figure 6.3 The mean recognition score for the girls and boys for the
two modes of task presentation
Figure 6.5 The mean ‘hits’ for the girls and boys in the test and game
conditions
The pattern of results thus suggests that the important distinction is not
between the beautician or technician contexts but rather between ‘game’
and ‘test’ contexts. If the context factor is treated as having only these two
levels, the resulting two-factor analysis of variance shows not only that
boys do rather better than girls overall, but more interestingly a signiWcant
interaction between gender and context (see Figure 6.5).
Further analysis reveals that where the task is introduced as a test there
is no trace of a performance diVerence between the girls and the boys. By
contrast, when exactly the same task is introduced as a game the boys’ and
girls’ performance diverges to a point where there is a signiWcant gender
diVerence in performance favouring the boys. Moreover, the girls per-
formed signiWcantly better when the task was described as a test than
when it was described as a game.
The data suggest that the gendered speciWcation of the ‘test’ contex-
tualisation (as beautician or technician) had no eVect on the performance
of girls and boys. More generally the boys’ performance appears to be
very little aVected by the context of the task presentation. This is not a
ceiling eVect, as there was considerable room for improvement. The girls’
Social comparison and learning 87
during lesson time, instructs them to play a computer game. What is the
purpose of such activity? The playing of computer games is not part and
parcel of recognised daily classroom activity. Thus there is a mismatch
between the girls’ understanding of what constitutes an appropriate
school task and the task they were asked to perform. We might thus
speculate that the girls do not know how to make sense of this interaction
with the experimenter, that they are unsure as to the purpose and mean-
ing of the activity they have been asked to undertake, since computer
gaming has nothing to do with the business of schooling.
Seen in these terms, then, the girls’ substantially better performance in
the ‘test’ conditions would reXect the fact that the intentions and motives
of the experimenter and the purpose of the activity is less ambiguous and
more appropriate to a school context. Although the unfamiliar adult is
asking them to perform a computerised vocational skills test, which they
would not be familiar with, a test is an activity wholly appropriate to the
classroom context. Tests are an integral feature of the school experience.
Moreover, adults who are not members of teaching staV (for example,
educational psychologists) can often be seen in school administering tests
to children.
This account rests on the assumption that girls tend to prefer ‘purpose-
ful’ activity and that they demonstrate sensitivity to context in a way that
boys tend not to. This is an assumption which has some grounding in the
wider educational literature. For example, data from the Assessment of
Performance Unit in Science reveal that within the domain of science,
girls take into account the circumstances within which a task is set,
whereas boys, as a group, consider the task in isolation from its context
(Murphy, 1993).
A second possible explanation, also predicated on the notion of girls’
sensitivity to context, concerns the cultural signiWcations of the context
‘game’. There is a vast amount of literature pointing to the signiWcant
association between gender and frequency of electronic game playing
(GriYths, 1991). Males tend to have considerably more computer-game
experience than females, both in childhood (Subrahmanyam and Green-
Weld, 1994) and adulthood (GreenWeld, Brannon and Lohr, 1994). They
have ‘learned how to learn’ video games and have played and practised on
them extensively and enthusiastically. Girls tend not to be enthusiastic
game players. For example, when interviewing children about their
thoughts on video games Stutz found that the girls were less keen than
boys on gaming because they saw the computer world as being biased
towards the male perspective and they were therefore discouraged from
taking an interest in them (Stutz, 1996). It is therefore possible that the
contextualising of the activity as a ‘game’ eVectively categorises the task as
Social comparison and learning 89
Overview
The results of these various studies show that, quite independently of the
intrinsic diYculty of the tasks concerned, presenting a task on a computer
can make a diVerence to performance. Not only that, but computer
presentation has a diVerential impact on the performance of girls and boys.
As the Wrst result in this section showed, this diVerential eVect is evident
even when the children do not need actually to do anything with the
computer at all. We have also seen that variations in the contextualisation
of the task also makes a diVerence as a function of gender. Thus we are
dealing here with eVects which owe more to individuals’ self-perceptions
and attributions than to their abilities – if indeed we can still meaningfully
separate these. The clear message emerging from this work is that there is
a fundamental interrelationship between the cognitive and the social; the
social and emotional dimensions of learning are not analytically separate
or distinct from the activity of learning. A learning situation is an inex-
tricably social situation. The recognition of this poses considerable
methodological and theoretical challenges. How best to study and con-
ceptualise the learner and the activity of learning? Does it make sense to,
as we have done in places, appeal to the notion ‘ability’ or refer to ‘high
and low achievers’ when we have seen that the children’s on-task per-
formance is contextually determined? In the Wnal chapter we attempt to
address some of these issues.
7 Interaction and learning: rethinking
the issues
101
102 References
Bone, A. (1983) Girls and Girls-only Schools: A Review of the Evidence, Man-
chester, Equal Opportunities Commission.
Bruner, J. (1985) ‘Vygotsky: A historical and conceptual perspective’ in
Wertsch, J. (ed.) Culture, Communication and Cognition: Vygotskian Per-
spectives, Cambridge, Cambridge University Press.
Bruner, J. (1996) The Culture of Education, Cambridge, MA, Harvard Univer-
sity Press.
Buckley, P. and Smith, P. (1991) ‘Opting out of technology: A study of girls’
GCSE choices’ in Lovegrove, G. and Segal, B. (eds.) Women into Com-
puting, Heidelberg, Springer-Verlag.
Burman, E. (1999) ‘Morality and the goals of development’ in Woodhead,
M., Faulkner, D. and Littleton, K. (eds.) Making Sense of Social Develop-
ment, London, Routledge, 170–80.
Burnham, W. (1905) ‘The group as a stimulus to mental activity’, Science, 31,
pp. 761–76.
Butler, R. and Ruzany, N. (1993) ‘Age and socialisation eVects on the
development of social comparison motives and normative ability assess-
ment in kibbutz and urban children’, Child Development, 64, pp. 532–43.
Butterworth, G. and Light, P. (1982) Social Cognition: Studies in the Develop-
ment of Understanding, Chicago, University of Chicago Press.
Camp, T. (1997) ‘The incredible shrinking pipeline’, Communications of the
Association for Computing Machinery (ACM), 40 (10), pp. 103–11.
Clements, D. (1987) ‘A longitudinal study of the eVects of Logo program-
ming on cognitive abilities and achievement’, Journal of Educational
Computing Research, 3 (1), pp 73–94.
Clements, D. and Nastasi, B. (1992) ‘The role of social interaction in the
development of higher order thinking in Logo environments’ in De
Corte, E., Linn, M., Mandl, H. and VerschaVel, L. (eds.) Computer-
Based Learning Environments and Problem Solving, Berlin, Springer-
Verlag.
Cole, M. (1996) Cultural Psychology: A Once and Future Discipline, Cam-
bridge, MA, The Belknap Press of Harvard University Press.
Cooper, J., Hall, J. and HuV, C. (1990) ‘Situational stress as a consequence
of sex-stereotyped software’, Personality and Social Psychology Bulletin,
16, pp. 419–29.
Costa, E. (1991). ‘The present and future of intelligent tutoring systems’ in
Scanlon, E. and O’Shea, T. (eds.) New Directions in Educational Technol-
ogy, Berlin, Springer-Verlag.
Cottrell, N., Wack, D., Sekerak G. and Rittle, R. (1968) ‘Social facilitation of
dominant responses by the presence of an audience and the mere pres-
ence of others’, Journal of Personality and Social Psychology, 47, pp.
1391–98.
Crocker, T. and Cheesman, R. (1988) ‘The ability of young children to rank
themselves for academic ability’, Educational Studies 14 (1), pp. 105–10.
Crook, C. (1987) ‘Computers in the classroom: DeWning a social context’ in
Rutkowska, J. and Crook, C. (eds.) Computers, Cognition and Develop-
ment, Chichester, Wiley.
References 103
116
Index 117
and gender: attitudes, 52, 66–7, 69–70, Gabrenya, W., 77
71; context, 81–9; performance, 56, Gallimore, R., 8, 9
58, 59, 61, 62, 65–6, 69–71; styles, games
53–4, 65–6; usage, 53, 88 gendered performance in context of,
group use of see collaborative learning 84–9
growing interest in, 26, 27 macho culture of, 53
mediating role of, 96–7 gender
concrete operations tasks, 3–4 attitudes to computers, 52, 66–7, 69–70,
conXict see socio-cognitive conXict 71, 88–9
conservation judgements peer interaction in learning, 54–5
Perret-Clermont’s work on, 4–5 performance with computers, 56, 58, 59,
problems of standard tests on, 5–6 61, 62, 65–6, 69–71; contextual
constructive approach, 2–5, 29–30 inXuences, 81–9
context software contrast studies, 60–4, 70–2
of collaboration, 93–6 stereotypes in software, 53, 58–60, 71–2
computer and manual, 81–4 style of computer use, 53–4, 65–6
disciplinary, 78–81 usage of computers, 53, 88
of performance, 97–9 see also mixed-sex groups; same-sex
contingent strategies, 9–10 groups
Cooper, J., 71 Gorsuch, C., 6
Cottrell, N., 75 groupwork
cultural context of collaboration, 93–6 move towards, xv
cultural learning, theories of, xiv–xv and use of computers, xv–xvi
cultural psychology, xvii see also collaborative learning; mixed-sex
see also socio-cultural approach groups; same-sex groups
cultural tools, 11–12
Hall, J., 71
decentration, 2–3 Hassan, R., 76
developmental psychology Hedley, J., 76
nature of, 99–100 history of experience, 93
social comparison in, 73 home computers, gendered use of, 53
disciplines, performance in context of, Honeybears studies
78–81 collaborative work on, 45–9, 50–1, 59
Doise, W., xvi, 3–4 features of, 43–4
Donaldson, M., 5 gender inXuences in, 58–9; peer
Durkin, K., 73 interaction, 65–6, 67, 70; software
contrasts, 60–4
education impact of peer presence, 49–50, 51
increasing use of computers in, xv–xvi, HuV, C., 71
27; role of psychology in, 27–31 Huguet, P., 74, 75–7, 78
educational software Humphreys, J., 83–4
contrast studies and gender, 60–4,
70–2 identity construction, link with thinking,
gender stereotypes in, 53, 58–60, 71–2 99
Edwards, D., 10 individual working see coactive working;
experimental research peer presence
context of, 95–6 instruction
see also computer tasks computer-assisted, 27, 28, 31
intelligent tutoring systems, 28
feedback, impact on performance, 74, in Tower of Hanoi studies, 16–17
76–7, 78 intellectual development, as social process,
Festinger, L., 73 xvi–xvii
Flett, G., 73 intelligent tutoring systems, 28
Forman, E., 11 interaction see peer interaction; social
friendship, impact on productivity, 24–5 interaction
118 Index
Joiner, R., 71 see also collaborative learning; mixed-sex
justiWcation, in balance beam studies, 23, groups; same-sex groups
24, 25–6 peer presence
impact on performance, 49–50, 51, 65,
King and Crown studies 74–6
collaborative work on, 36–9, 40–1, 42, see also coactive working
50 peer tutors, 10
features of, 33–6, 40 performance
gender inXuences in, 56–8; software impact of computer context on, 81–4
contrasts, 60–1, 63–4 impact of discipline context on, 78–81
knowledge, social construction of, 92 impact of friendship on, 24–5
impact of peer presence on, 49–50, 51,
Latane, B., 77 74–6
learning impact of social comparison on, 74–8
changing notions of, 92 relationship with gender, 56, 58, 59, 61,
contingent strategies in, 9–10 62, 65–6, 69–71
dimensions of, 99 as situated, 97–9
and social comparison, 77–84 Perret-Clermont, A.-N., 4–5, 7
Vygotskian approach to interaction in, Piaget, J., xvi, 2–3
8–11 Pirates studies, 61–4
see also collaborative learning Princesses study, 67
Light, P., 5–6, 71 productivity see performance
Littleton, K., 71 programming computers, 29, 31
Logo, 29, 31 psychology of learning, impact on
computer use, 27–31
Malone, T., 89
Manstead, A., 75 Resnick, L., 92, 93
Marshall, H., 74 Ringelmann, M., 76
Mash, E., 76 Rittle, R., 75
Mastermind studies, 20–2, 25 Robbins, A., 5–6
maths, performance in context of, 78–81 Robinson-Staveley, K., 71
mediated action, 11–12 role division, in King and Crown studies,
memory tests, gender and context of, 39
78–83 Roschelle, J., 47
Mercer, N., 10 Ross, G., 9
Messer, D., 71 Ruble, D., 73
mixed-sex groups working with computers, Russell, J., 5
54–5
Honeybears studies, 59 same-sex groups working with computers,
interaction and coaction, 64–70, 71 54, 55
King and Crown studies, 56–8 Honeybears studies, 59
modelling, 1, 17, 28 interaction and coaction, 64–70, 71
Monteil, J.-M., 74, 75–8 King and Crown studies, 56–8
mouse control, and gender, 58, 59 Sanna, L., 76–7
scaVolding, 9, 10
Newman, J., 6 school, girls’ understanding of, 87–8
Sekerak, G., 75
operational modes of thought, 2–4 Semin, G., 75
Seta, J., 76
Papert, S., 29 simulated worlds, 29–30
peer interaction single-sex groups see same-sex groups
and gender in learning, 54–5 situated learning approach, xvii
impact on development, 2–5, 9, 10–11 see also context
impact of friendship on, 24–5 Skinner, B. F., 27
on social class diVerences, 7 social class diVerences, 7
Index 119
social comparison, 73–7 Tharp, R., 8, 9
and children’s learning, 77–84 thinking
social construction link with identity construction, 99
of knowledge, 92 see also cognition
see also constructivist approach tools, 11–12, 96
social interaction see also computers
Vygotskian approach to learning and, Tower of Hanoi studies, 14–20, 25, 83–4
8–11 Triplett, N., 74
see also peer interaction tutoring see instruction
social learning theory, 1–2
social loaWng, 76–7 unstructured interaction, 17, 18
social-cognition approach, xiii–xiv
socio-cognitive conXict, 4, 17, 26, 91 verbal interaction, 54, 58
reservations about, 5–7 in Honeybears studies, 48
socio-cultural approach, 11–12 in King and Crown studies, 40–1
software in Mastermind studies, 21–2, 25
contrast studies and gender, 60–4, 70–2 in Tower of Hanoi studies, 19–20,
gender stereotypes in, 53, 58–60, 71–2 25
Steele, F., 71 verbal justiWcation, in balance beam
Stroop test, 75 studies, 23, 24, 25–6
structured interaction, 17, 18, 19 village task, 3–4
style of computer use, 53–4, 65–6 Vygotsky, L. S., xvi–xvii, 8–9, 11
symbolic tools, 11–12 inXuence of, 7–8, 9–11
symmetry, in collaborative working, 45–8
Wack, D., 75
talk see verbal interaction Wang, Y., 77
teaching Weinstein, R., 74
contingent strategies in, 9–10 Wertsch, J., 11–12
ZPD in theory of, 8 Wood, D., 9–10
Teasley, S., 47
test, gendered performance in context of, Zajonc, R., 74–5
84–9 zone of proximal development (ZPD), 8–9